import os
import cv2
import time
from PIL import Image

#只實(shí)現(xiàn)截屏的功能
global path
path='./images/'
#人臉采樣,封裝函數(shù)
def cy(path):
   #path為保存圖片的路徑
   #調(diào)用筆記本內(nèi)置攝像頭，參數(shù)為0，如果有其他的攝像頭可以調(diào)整參數(shù)為1,2
   cap = cv2.VideoCapture(0)
   #為即將錄入的臉標(biāo)記一個(gè)id
   face_id = input('\n 用戶臉部信息錄入，輸入用戶名字（最好用英文）：\n')
   #sampleNum用來(lái)計(jì)數(shù)樣本數(shù)目
   count = 0
   
   while True:
       #從攝像頭讀取圖片
       success,img = cap.read()    
       count += 1  
       #保存圖像，把灰度圖片看成二維數(shù)組來(lái)檢測(cè)人臉區(qū)域
       #保存到相應(yīng)的文件夾里
       cv2.imwrite(path+str(face_id)+'.'+str(count)+'.jpg',img)
       #顯示圖片
       cv2.imshow('image',img)
           #保持畫(huà)面的連續(xù)。waitkey方法可以綁定按鍵保證畫(huà)面的收放，通過(guò)q鍵退出攝像
       k = cv2.waitKey(1)
       if k == '27':
           break
           #或者得到800個(gè)樣本后退出攝像，這里可以根據(jù)實(shí)際情況修改數(shù)據(jù)量，實(shí)際測(cè)試后800張的效果是比較理想的
       elif count >= 500:
           time.sleep(2)
           success,img = cap.read()
           break
   
   #關(guān)閉攝像頭，釋放資源
   cap.release()
   cv2.destroyAllWindows()
#調(diào)用函數(shù)進(jìn)行人臉采樣
cy(path)

import cv2
import os

#對(duì)圖片進(jìn)行處理，輸入的不是灰度圖片,方便外部采集的圖片進(jìn)行處理
CASE_PATH = "haarcascade_frontalface_default.xml"
RAW_IMAGE_DIR = 'images/'
DATASET_DIR = 'hh/'
path='D:\\pythonlx\\test\\images\\'

#人臉?lè)诸惼?
face_cascade = cv2.CascadeClassifier(CASE_PATH)

#定義人臉大小
def save_feces(img, name,x, y, width, height):
   image = img[y:y+height, x:x+width]
   cv2.imwrite(name, image)

image_list = os.listdir(RAW_IMAGE_DIR) #列出文件夾下所有的目錄與文件
for image_path in range(len(image_list)):
   gh=path+image_list[image_path]
  # print(gh)
   image = cv2.imread(gh)
  #gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
   faces = face_cascade.detectMultiScale(image,
                                        scaleFactor=1.2,
                                        minNeighbors=5,
                                        minSize=(5, 5), )
   for (x, y, width, height) in faces:
       save_feces(image, '%ss%d.jpg' % (DATASET_DIR, image_path+1), x, y - 30, width, height+30)

 #讀取圖片
def read_image():
 data_x, data_y = [], []
 image_list = os.listdir('mine/')
 for i in range(len(image_list)):
     try:
         im = cv2.imread('mine/{}'.format(image_list[i]))
         im = resize_without_deformation(im)
         data_x.append(np.asarray(im, dtype = np.int8))
         #定義標(biāo)簽
         a=image_list[i].split('.')[0]
         if a=='s2':
             data_y.append(0)
         elif a=='s4':
             data_y.append(1)
         elif a=='s5':
             data_y.append(2)
         elif a=='s6':
             data_y.append(3)
         elif a=='s7':
             data_y.append(4)
     except IOError as e:
         print(e)
     except:
         print('Unknown Error!')
 return data_x,data_y

#讀取所有圖片以及標(biāo)簽
raw_images, raw_labels = read_image()
##查看數(shù)據(jù)每個(gè)標(biāo)簽的數(shù)據(jù)量
#a=raw_labels.count(0)#583
#b=raw_labels.count(1)#621
#c=raw_labels.count(2)#717
#d=raw_labels.count(3)#765
#e=raw_labels.count(4)#698

#轉(zhuǎn)為浮點(diǎn)型
raw_images, raw_labels = np.asarray(raw_images, dtype = np.float32),np.asarray(raw_labels, dtype = np.int32)

#將標(biāo)簽轉(zhuǎn)化為one_hot類型
ont_hot_labels = np_utils.to_categorical(raw_labels)

#劃分?jǐn)?shù)據(jù)集，訓(xùn)練集80%，測(cè)試集20%
train_input, valid_input, train_output, valid_output =train_test_split(raw_images, 
               ont_hot_labels,
               test_size = 0.2)
#數(shù)據(jù)歸一化處理
train_input /= 255.0
valid_input /= 255.0

def resize_without_deformation(image, size = (100, 100)):
  height, width, _ = image.shape
  #對(duì)于長(zhǎng)度不等的邊，找到最長(zhǎng)邊
  longest_edge = max(height, width)
  #使用0填充邊框
  top, bottom, left, right = 0, 0, 0, 0
   #計(jì)算短邊需要增加多上像素寬度使其與長(zhǎng)邊等長(zhǎng)
  if height < longest_edge:
      height_diff = longest_edge - height
      top = int(height_diff // 2)
      bottom = height_diff - top
  elif width < longest_edge:
      width_diff = longest_edge - width
      left = int(width_diff // 2)
      right = width_diff - left
   #給圖像增加邊界，是圖片長(zhǎng)、寬等長(zhǎng)，cv2.BORDER_CONSTANT指定邊界顏色由value指定
  image_with_border = cv2.copyMakeBorder(image, top , bottom, left, right, cv2.BORDER_CONSTANT, value = [0, 0, 0])

  resized_image = cv2.resize(image_with_border, size)
  #將裁剪的圖片轉(zhuǎn)化為灰度圖
  resize_image= cv2.cvtColor(resized_image,cv2.COLOR_BGR2GRAY)
  #直方圖均衡化
  hist = cv2.equalizeHist(resize_image)
  img2 = hist.reshape((100, 100, 1))
  return img2

 #搭建卷積神經(jīng)網(wǎng)絡(luò)，順序模型
model = models.Sequential()

#卷積層，卷積核大小32，每個(gè)大小383,步長(zhǎng)為1，輸入的類型為（100,100,1）,1為通道,激活函數(shù)relu
model.add(Conv2D(filters=32,kernel_size=(3,3),padding='valid',strides= (1, 1),#1
                                  input_shape = (100, 100,1),
                                  activation='relu'))

model.add(Conv2D(filters=32,kernel_size=(3,3),padding='valid',strides= (1, 1),#2
                                  activation='relu'))


#池化層
model.add(MaxPooling2D(pool_size=(2, 2)))#3

#Dropout層
model.add(Dropout(0.25))#4

#卷積層
model.add(Conv2D(64, (3, 3), padding='valid',
                                  strides = (1, 1),
                                  activation = 'relu'))#5

model.add(Conv2D(64, (3, 3), padding='valid',
                                  strides = (1, 1),
                                  activation = 'relu'))#6

#池化層
model.add(MaxPooling2D(pool_size=(2, 2)))#7
model.add(Dropout(0.25))#8

#全連接
model.add(Flatten())#9

model.add(Dense(512, activation = 'relu'))#10

model.add(Dropout(0.25))#11

#輸出層，神經(jīng)元數(shù)是標(biāo)簽種類數(shù)，使用sigmoid激活函數(shù)，輸出最終結(jié)果
model.add(Dense(len(ont_hot_labels[0]), activation = 'sigmoid'))#12



#優(yōu)化模型,
#SGD----梯度下降算子
#learning_rate = 1#學(xué)習(xí)率
#decay = 1e-6#學(xué)習(xí)率衰減因子
#momentum = 0.8#沖量
#nesterov = True
#sgd_optimizer = SGD(lr = learning_rate, decay = decay,
#                    momentum = momentum, nesterov = nesterov)

#categorical_crossentropy
#優(yōu)化器用Adam算法，損失函數(shù)用交叉熵的方法
model.compile(optimizer='adam',loss='categorical_crossentropy',
              metrics=['accuracy'])


#輸出模型參數(shù)
model.summary()

#定義數(shù)據(jù)生成器用于數(shù)據(jù)提升，其返回一個(gè)生成器對(duì)象datagen，datagen每被調(diào)用一
#次其生成一組數(shù)據(jù)（順序生成），節(jié)省內(nèi)存，其實(shí)就是python的數(shù)據(jù)生成器
datagen = ImageDataGenerator(
                 featurewise_center = False,             #是否使輸入數(shù)據(jù)去中心化（均值為0），
                 samplewise_center  = False,             #是否使輸入數(shù)據(jù)的每個(gè)樣本均值為0
                 featurewise_std_normalization = False,  #是否數(shù)據(jù)標(biāo)準(zhǔn)化（輸入數(shù)據(jù)除以數(shù)據(jù)集的標(biāo)準(zhǔn)差）
                 samplewise_std_normalization  = False,  #是否將每個(gè)樣本數(shù)據(jù)除以自身的標(biāo)準(zhǔn)差
                 zca_whitening = False,                  #是否對(duì)輸入數(shù)據(jù)施以ZCA白化
                 rotation_range = 20,                    #數(shù)據(jù)提升時(shí)圖片隨機(jī)轉(zhuǎn)動(dòng)的角度(范圍為0～180)
                 width_shift_range  = 0.2,               #數(shù)據(jù)提升時(shí)圖片水平偏移的幅度（單位為圖片寬度的占比，0~1之間的浮點(diǎn)數(shù)）
                 height_shift_range = 0.2,               #同上，只不過(guò)這里是垂直
                horizontal_flip = True,                 #是否進(jìn)行隨機(jī)水平翻轉(zhuǎn)
                 vertical_flip = False)                  #是否進(jìn)行隨機(jī)垂直翻轉(zhuǎn)
 
#計(jì)算整個(gè)訓(xùn)練樣本集的數(shù)量以用于特征值歸一化、ZCA白化等處理
datagen.fit(train_input)

#利用生成器開(kāi)始訓(xùn)練模型
history=model.fit_generator(datagen.flow(train_input, train_output,batch_size = 50),
                                      epochs = 10,
                                      validation_data = (valid_input, valid_output))

#模型驗(yàn)證
print(model.evaluate(valid_input, valid_output, verbose=2))

##畫(huà)出LOSS變化曲線圖
plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'],label='val_accuracy')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.ylim([0.5,1])
plt.legend(loc='lower right')

#保存模型
MODEL_PATH = 'face_model.h5'
model.save(MODEL_PATH)